Myers and Zamudio, Page 1
Multiple Paternity in an Aggregate Breeding Amphibian: the Effect of1
Reproductive Skew on Estimates of Male Reproductive Success2
3
E. M. MYERS* and K. R. ZAMUDIO4
Dept. of Ecology and Evolutionary Biology, Cornell University, Ithaca, NY 14853, USA5
* Present address: Dept. of Ecology, Evolution, and Organismal Biology, Iowa State6
University, Ames, IA 50011, USA7
8
Date of Receipt:9
Keywords: microsatellites, multiple paternity, reproductive skew, Ambystoma, mating10
system11
12
13
Running title: salamander multiple paternity14
Myers and Zamudio, Page 2
Abstract15
Aggregate or explosive breeding is widespread among vertebrates and likely16
increases the probability of multiple paternity. We assessed paternity in seven field17
collected clutches of the explosively breeding spotted salamander (Ambystoma maculatum)18
using ten microsatellite loci to determine the frequency of multiple paternity and the19
number of males contributing to a female’s clutch. Using the Minimum Method of allele20
counts, multiple paternity was evident in 70% of these egg masses. Simple allele counts21
underestimate the number of contributing males because this method cannot distinguish22
multiple fathers with common or similar alleles. Therefore, we used computer simulations23
to estimate from the offspring genotypes the most likely number of contributing fathers24
given the distributions of allele frequencies in this population. We determined that two to25
eight males may contribute to A. maculatum clutches; therefore, multiple paternity is a26
common strategy in this aggregate breeding species. In aggregate mating systems27
competition for mates can be intense, thus, differential reproductive success (reproductive28
skew) among males contributing to a female’s clutch could be a likely outcome. We use29
our data to evaluate the potential effect of reproductive skew on estimates of the number of30
contributing males. We simulated varying scenarios of differential male reproductive31
success ranging from equal contribution to high reproductive skew among contributing32
sires in multiply-sired clutches. Our data suggest that even intermediate levels of33
reproductive skew substantially decrease confidence in estimates of the number of34
contributing sires when parental genotypes are unknown.35
Myers and Zamudio, Page 3
Introduction36
Reproductive strategies in vertebrates fall along a continuum depending on the37
degree of monopolization of mates, availability of resources, and duration of courtship38
(Emlen & Oring 1977; Wells 1977). At one end of this continuum is territorial monogamy,39
a system in which males defend areas from other males and/or guard females with whom40
they mate; in these systems mating season and courtship usually take place over several41
weeks or months (Wells 1977). At the other end of this spectrum is aggregate, or42
explosive, breeding that occurs over a very short time period with little or no43
monopolization of either space or individuals (Wells 1977). In aggregate mating systems,44
males and females typically arrive together at a neutral location for a brief, intense bout of45
mating. Under these circumstances a ‘lottery system’ is established where each competing46
individual has a very small opportunity to mate, but few are actually successful, therefore47
individuals vary dramatically in realized reproductive success (Nunney 1993).48
Aggregate breeding is widespread across many vertebrate taxa including fish,49
amphibians, and reptiles (including birds) (Roberts 1994; Willmott & Foster 1995;50
Birkhead 2000). Aggregate breeding may have evolved in response to a number of51
selective environments. For example, selection to breed in a narrow time window when52
environmental conditions are optimal, or selection to reduce time and energy costs53
associated with pair bonding may have favored the evolution and maintenance of a system54
where courtship is rapid and limited. Regardless of the selective pressures promoting its55
evolution, aggregate breeding always allows for large amounts of genetic mixing to take56
place in a short time because it presents an opportunity for both males and females to mate57
multiply and hence produce offspring with different genetic compositions. If genetic58
Myers and Zamudio, Page 4
mixing is an advantage of this mating system, we expect that aggregate-breeding taxa will59
generally exhibit multiple paternity. Although we have estimates for patterns of male60
reproductive success in vertebrate mating systems, particularly in birds and mammals61
(MacWhirter 1991; Roff 2002), we know significantly less about reproductive patterns in62
amphibians, despite their high diversity of mating systems and reproductive modes63
(Haddad and Pombal 1998; Denoel et al. 2001). For example, despite ample opportunity64
for multiple paternity in many amphibian mating systems (Roberts et al. 1999), few cases65
of multiple paternity have been documented in frogs and salamanders (but see Jones et al.66
2002), and the mating system of the more cryptic caecilians is virtually unknown.67
Determination of paternity in amphibians (and ectotherms in general) has been68
challenging due to limitations imposed by mating behavior and life history traits specific to69
these lineages. Many amphibians and fishes have very large clutch or litter sizes, which70
have necessitated the development of statistical and methodological methods for analyzing71
subsamples of large broods (DeWoody et al. 2000). Likewise, most ectotherms show72
limited parental care or association with their young, and it is not always possible to73
sample or fingerprint potential parents, prompting the development of simulation74
techniques for inference of contributing parents or reconstruction of half sib groups75
(Blouin ref, DeWoody et al. 2000; Fiumera et al. 2001). Finally, in multiply-sired76
clutches, unequal contribution by sires (reproductive skew) can affect paternity estimates,77
although the degree to which reproductive skew biases estimates of the number of78
contributing parents has not been rigorously evaluated.79
The spotted salamander, Ambystoma maculatum, is an aggregate-breeding80
amphibian, and thus an ideal species in which to characterize reproductive success and81
Myers and Zamudio, Page 5
multiple paternity as the first step in understanding the fitness consequences of this mating82
system. Adult spotted salamanders migrate to vernal pools in the early spring for short,83
ardent, 1- 3 day bouts of breeding (Hillis 1977; Petranka 1998), an extremely limited time84
for interaction between the sexes. In Ambystoma, fertilization is internal although sperm85
transfer is indirect via spermatophores (sperm packets atop a gelatinous base) that males86
deposit on the pond floor (Halliday and Verrell 1984). Individuals of both sexes87
congregate in large numbers, swimming, nudging and rubbing each other (Bishop 1941;88
Petranka 1998). Males deposit, on average, 20-40 spermatophores on the pond bottom and89
females pick up 15-20 of these in one breeding night (Arnold 1976; Petranka 1998).90
Female spotted salamanders search for spermatophores with their hindlimbs while moving91
along the substrate; upon encounter, a female picks up the sperm cap with her cloaca92
leaving behind the gelatinous base, thus gaining no benefits aside from the sperm (Arnold93
1976; Petranka 1998). Males do not limit female access to spermatophores making it94
possible for a female to be inseminated by multiple males in the courting group. Shortly95
after mating, females lay their eggs on submerged vegetation in the pond (Petranka 1998),96
where larvae hatch and develop with no parental care.97
Given our current knowledge of natural history and breeding behavior of this98
species we have several expectations for the reproductive system of Ambystoma99
maculatum. We predict that an aggregate-explosive breeder with multiple matings, such as100
spotted salamanders (Arnold 1976), should exhibit high levels of multiple paternity in101
nature. In this study, we conducted a paternity analysis to genetically characterize the102
reproductive system in this species by assessing paternity of field-collected egg masses.103
Using computer simulations that take into account the genetic structure of the focal104
Myers and Zamudio, Page 6
population, and assuming an equal contribution of sires to a female’s clutch, we105
determined the minimum sample size of eggs in a clutch needed for detecting multiple106
sires and estimated the number of males that contribute to clutches using both direct107
genotype counts and computer simulations. Given the intensity of male-male competition,108
if multiple paternity is common it is possible that the contribution of males to offspring109
within a clutch will be skewed. We use our dataset to quantify the potential impact of110
reproductive skew on estimates of paternity. We simulate a range of skew scenarios (from111
even paternal contribution to highly skewed clutches) and compare the estimates of112
number of contributing males and their statistical confidence.113
Characterizing the degree of multiple paternity and reproductive fitness in this114
system increases our understanding of the evolution and maintenance of aggregate mating115
systems. Likewise, quantifying the potential impact of reproductive skew on male paternity116
estimates will help guide development of future theoretical approaches for quantifying117
reproductive success in systems where parental alleles are not known as well as the design118
and analysis of empirical data that address reproductive success in other systems where119
reproductive skew is likely to occur.120
121
Materials & Methods122
Field site and sample collections123
We collected seven spotted salamander egg masses from Ringwood Pond,124
Tompkins County, New York, USA during the 2000 and 2001 breeding seasons. Clutches125
A, B, and C were collected in spring 2000; clutches D, E, F, and G were collected in 2001.126
All egg masses were collected three to four days after the commencement of the breeding127
Myers and Zamudio, Page 7
season and thus likely within 24 hours of oviposition. Egg masses were brought back to the128
lab and maintained in individual containers (500 or 2000 ml) with dechlorinated water.129
Clutches were monitored daily for development; larvae were allowed to hatch naturally130
and were then preserved for later genetic analyses. Original clutch sizes ranged from 30 -131
200 eggs and numbers of hatched larvae from these clutches ranged from 7-51. Consistent132
with previous field and lab studies, egg mortality was approximately 50-90% of each133
clutch, and was assumed to affect eggs with equal probability (Stenhouse 1987; Metts134
2001; Tennessen & Zamudio 2003). All preserved larvae were used in subsequent135
paternity analyses.136
137
Genotyping138
Whole preserved larvae were digested using Proteinase K (100 µg) in 500 µL lysis139
buffer (2.5 mM Tris, 8 mM NaCl, 0.1%SDS, 0.2 mM EDTA, 0.01% ß-mercapto-ethanol).140
Samples collected in 2001 were additionally treated with 10 mg/mL RNAseA. Genomic141
DNA was purified using standard phenol-chloroform organic cleanup and ethanol142
precipitation protocols (Sambrook & Russell 2001). Extracts were diluted in distilled water143
to a concentration of 100 ng/µl for use as templates for amplification of ten microsatellite144
loci via the polymerase chain reaction (PCR) (Wieczorek et al. 2002). Amplified products145
were combined in multiplex groups of 3 or 4 loci, and electrophoresed on 5%146
polyacrylamide gels on an ABI prism 377 DNA Sequencer (Applied Biosystems , Costa147
Mesa, CA). Peaks were sized using TAMRA 350 or 500 size standards. Genotype results148
were analyzed using Genescan v.3.1 and Genotyper v.2.5 (Applied Biosystems, Costa149
Myers and Zamudio, Page 8
Mesa, CA). Fragment sizes were checked and assigned to pre-defined allele size-classes by150
hand (Wieczorek et al. 2002).151
152
Multiple Paternity Estimation153
Multiple paternity was assessed using two estimators: 1) the Minimum Method and154
2) a simulation estimate. The Minimum Method is more conservative and relies entirely on155
counts of alleles represented within clutches. The Minimum Method assigns multiple156
paternity within a clutch by using patterns of allelic distribution in offspring to determine157
maternal genotypes and assumes that all alleles not accounted for by maternal genotype158
were necessarily contributed by fathers. Therefore, a half-sib brood is evident if there are159
more than four alleles present at a given locus, two of which are maternal and three or160
more paternally contributed alleles. Multiple paternity is defined only in those cases in161
which 1) a maternal genotype could be reconstructed and three or more alleles were162
additionally present or 2) if maternal genotype could not be reconstructed but five or more163
alleles were present.164
The Minimum Method, although informative, is conservative in that it does not165
account for multiple paternity by males with similar genotypes. We used the COUNTS,166
BROOD, and HAPLOTYPES computer programs (Fiumera et al. 2001, DeWoody et al.167
2000) to further analyze our data. These programs overcome the difficulties of assessing168
exactly how many parents contributed to a progeny array in the absence of information on169
parental genotypes by taking into account allele frequencies in the adult population.170
BROOD determines the mean number of offspring that must be sampled to detect multiple171
paternity in a clutch, COUNT estimates the number of parental haplotypes from the172
Myers and Zamudio, Page 9
empirical genotype data, and HAPLOTYPES uses that information to estimate the number173
of contributing males given allele frequencies in the population.174
We used the BROOD program to determine the mean number of offspring175
necessary to detect all parental gametes in a brood. This program determines the mean176
number of offspring that must be sampled (along with 95% confidence intervals) to177
observe the same allele frequencies that are seen in the entire population and the number178
needed to detect every gamete in the parental gene pool, including those which may be179
shared amongst parents (N*).180
COUNTS determined the number of unique haplotypes of paternal origin evident in181
the offspring genotype data. This program deduces the alleles contributed by a male182
through subtraction, any allele in the progeny not displayed by the mother must have come183
from a father (Fiumera et al. 2001). We used the COUNTS_MED program; in the case of184
an offspring with the same heterozygous genotype as the mother, this program randomly185
assigns one of the two possible alleles as the presumed contribution from a sire and tallies186
the contribution only if it is unique (Fiumera et al. 2001). The number of haplotypes within187
the clutch is then used to determine an expected number of unique fathers.188
HAPLOTYPES simulates a breeding population based on empirical allele189
frequencies in the study population (DeWoody et al. 2000). We used HAPLOTYPES to190
simulate clutches that approximated our field collected egg masses in size and sampling191
and to assess the number of haplotypes one would expect to find in clutches sired by192
different numbers of males. Using the number of haplotypes determined from COUNTS193
for each clutch, we then estimated the number of males that most frequently produced that194
number of haplotypes in our simulated clutches.195
Myers and Zamudio, Page 10
We varied the parameters in the programs to simulate an array of conditions that196
may exist in spotted salamander aggregations. We simulated clutch sizes of 50 and 100197
eggs, consistent with field-collected clutch sizes that ranged from approximately 30-150198
eggs each. We also varied the number of males potentially contributing to a clutch from199
one to eight for BROOD and HAPLOTYPES. This range allows for single paternity and200
encompasses previous estimates of multiple paternity from controlled mating experiments201
(Tennessen & Zamudio 2003). The adult population at our study site was estimated at 1700202
individuals (Zamudio, unpublished) and both parents and progeny were sampled 1000203
times in all simulations. For the initial estimate of male reproductive success we assumed204
equal contributions from each of the males contributing to a clutch (i.e. no within clutch205
reproductive skew).206
Both BROOD and HAPLOTYPES simulate breeding populations based on207
empirical allele frequencies. Allele frequencies for each of the 10 microsatellite loci were208
estimated from two more extensive microsatellite studies in our focal population and209
surrounding ponds (Wieczorek & Zamudio, in prep; Tennessen & Zamudio 2003). Overall210
frequencies were estimated by pooling genotypes for 118 individuals for 5 neighboring211
ponds in the Ringwood watershed. These populations are panmictic (Wieczorek &212
Zamudio, in prep) and represent a single breeding deme. For most loci, the alleles included213
in our egg clutches were a subset of the pool of alleles found among adults in this breeding214
deme; in those cases we used the allele frequencies from the previous studies to directly215
estimate allele frequencies for our simulations. These loci have an expected heterozygosity216
that ranges from 0.329 to 0.820 (Wieczorek et al. 2002). At loci Ama 5-1, Ama 9-4, Ama217
12-7, Ama 34, and Ama A, a single allele that had not been detected in the original218
Myers and Zamudio, Page 11
watershed data set was represented in our clutches. For locus Ama 5-1 the rare allele was219
also evident in a previous study assessing paternity at Ringwood (Tennessen & Zamudio220
2003). To account for this allele, genotypes of 78 adults from Ringwood Pond (Tennessen221
& Zamudio 2003) were combined with the more extensive population data from the222
Ringwood watershed to recalculate overall allele frequencies for that locus. For the223
remaining four loci, the novel allele we detected in our clutches had not been present in224
either of the earlier studies. In these cases, we assigned an arbitrary but small allele225
frequency of 0.002 to the rare allele and reduced the allele frequency of the most common226
allele by the same margin. This value is the expected frequency of an allele seen only once227
in the Ringwood watershed data set, therefore, this recalculation did not dramatically228
change allele frequencies.229
230
Null Alleles231
A null allele is an allele that fails to amplify during PCR due to large increases in232
the size of the product or mutations in the flanking primer sites (Dowling et al. 1996). The233
presence of a null allele is indicated by genotype frequencies that do not conform to234
Mendelian expectations due to an overabundance of an apparent shared allele (Dowling et235
al. 1996; Fitzsimmons 1998). A maternal null allele was presumed present when three or236
more different homozygous offspring were observed within a clutch. Offspring receiving237
the null allele from the mother would appear homozygous for the paternal allele resulting238
in an apparent overabundance of homozygous individuals. Using the Minimum Method,239
we assigned “conditional multiple paternity” in situations where a null allele was240
determined to be present as part of the maternal genotype and four additional alleles were241
Myers and Zamudio, Page 12
present among the offspring. None of the computer programs permit simulations of the242
occurrence of null alleles in the population, and the COUNTS program does not accept loci243
with null alleles. Thus, loci in which null alleles were evident were eliminated from244
subsequent analyses for that particular clutch.245
246
Quantifying the effects of reproductive skew on estimates of multiple paternity247
The original versions of BROOD and HAPLOTYPES only permit simulation of248
clutches with equal male representation within clutches, thus we modified those programs249
to permit simulations using different parameters that characterize reproductive skew.250
Modified programs are available from the first author (E.M.M) upon request. We251
simulated male reproductive skew within clutches in three ways. First, reproductive252
success among fathers is assumed to follow a geometric distribution based on the253
following equation254
Proportion of offspring from jth father=a (1-a)j-1255
where a is a measure of increasing reproductive skew. Under this scenario of skew, the256
most successful male sires a fraction (a) of the offspring in a females clutch, and each257
successive male sires the same proportion of the remaining eggs. We simulated scenarios258
of skew where a was equal to 0.125 (where eight males sire even proportions of remaining259
offspring in a clutch), and where a ranged from 0.2 to 0.9 at 0.1 intervals. This geometric260
distribution of paternity could be expected in cases where male precedence influences261
paternity rates, such that later males have smaller and smaller reproductive success. In the262
other two skew analyses we assumed that one dominant male fathered most of the263
offspring of the larvae in a clutch and the remaining larvae are distributed equally among264
Myers and Zamudio, Page 13
remaining males. In the first of these, the most successful (dominant) male sires twice as265
many offspring in the clutch than the other male or males (hereafter referred to as the 2X266
scenario). Finally we also simulated the scenario where the dominant male fathers 60% of267
all offspring and the remaining offspring are shared equally among the other males268
(hereafter referred to as the 60% scenario). Results from each simulation were compared to269
the simulations with no reproductive skew (i.e. all males contribute equally to clutches in270
which they sire offspring). Combined these three scenarios of reproductive skew cover a271
wide range of possibilities, some of which have been observed in previous studies focusing272
on determinants of fitness among competing males (Jones et al. 2002; Tennessen &273
Zamudio 2003).274
275
Results276
Minimum Method277
We genotyped 175 larvae from seven clutches at ten polymorphic loci and used278
these data to reconstruct patterns of allelic variation and paternity (Table 1). Using the279
Minimum Method, multiple paternity was observed in five out of seven clutches (clutches280
A, B, C, F and G). Multiple paternity was assigned unambiguously for at least one locus in281
clutches A, B, C, and F. Conditional multiple paternity (when there was evidence of a null282
allele) was evident in one locus of clutch G as well as in each of the other four clutches.283
All loci were used to determine the number of contributing fathers using Minimum284
Method; for clutches A, B, C, F, and G the minimum number of fathers represented was 3,285
3, 2, 3, and 2, respectively (Table 1).286
Myers and Zamudio, Page 14
In multiply sired clutches, male genetic contribution was not shared equally. In287
Table 1, loci highlighted as indicative of multiple paternity exhibit high frequencies of the288
maternal alleles and paternal alleles from the most common sires. The remaining alleles,289
contributed by other males, are lower in frequency in the clutch. Closer examination of the290
most frequent alleles within a clutch and comparison with population allele frequencies291
yields two results. In several of the clutches (e.g. Clutch C and E at locus Ama 61, Clutch292
A at locus Ama 11-2B) the most frequent allele within the clutch was also one of the most293
frequent alleles within the population (Table 1 and Appendix). This suggests that either 1)294
reproductive skew is present and the most successful male coincidentally had a common295
allele, or 2) skew is not high and several males with the most common allele contributed to296
these clutches and are categorized as a single male by the Minimum Method. In other297
instances the most common paternal allele in the clutch is relatively infrequent in the298
population and is more likely to represent a contribution from a single male, and thus more299
strongly supports high reproductive skew. For example, in clutch C, the most frequent300
allele at locus Ama 4-10 is allele 241, because this allele is relatively rare in the population301
with a frequency of only 0.052, it was most likely the contribution of one very successful302
male. Although these data do not allow us to quantify the degree of reproductive skew,303
they do suggest that not all males are contributing equally to egg clutches.304
Multiple maternity was found in a single clutch (clutch E). Two loci, Ama 5-1 and305
Ama 34 have offspring genotypes that are inconsistent with a single maternal genotype306
(Table 1). In addition, this also exhibits a much higher frequency of loci with null alleles307
(classified by three or more homozygous types) than the remaining clutches (0.7 compared308
to an average of 0.283 for the others). These excess homozygote larvae are likely the result309
Myers and Zamudio, Page 15
of the second mother and her mate(s). Clutch E was an unusually large clutch with310
approximately 175 eggs, significantly larger than the average clutch size in this region311
(Bishop 1941). It is likely that two females deposited eggs on the same branch and the jelly312
masses surrounding the eggs coalesced, making the egg masses appear as one clutch. This313
clutch was eliminated from subsequent analyses. Additionally, three clutches showed314
evidence of multiple maternity at loci Ama 5-1 (G and D) and Ama 34 (C). These loci315
were highly polymorphic and the extra maternal allele was present in very few offspring.316
In these cases we assumed that the additional maternal allele was not due to multiple317
maternity, but resulted from a single mutation or genotyping error because the rare allele318
appears in only 1-3 larvae. These clutches were retained for subsequent analyses (although319
the individual loci were not).320
Single paternity was found in a single clutch (clutch D). At all but one locus (Ama321
5-1, discussed above) this clutch exhibited a single male genotype pattern. This clutch was322
one of the smallest clutch included in the study and would be unlikely to demonstrate323
multiple paternity based on allele counts.324
325
Null Alleles326
Null alleles were relatively common in our data producing a string of homozygous327
offspring exhibiting alleles from only one parent (Figure 1). In analyzing these results one328
can maximize the number of observed alleles by including all alleles irrespective of the329
frequency with which they are seen. Alternatively, a more conservative approach excludes330
alleles only seen once, under the assumption that they result from mutation or genotyping331
error. This method necessarily produces lower numbers of null alleles; however, it also332
Myers and Zamudio, Page 16
decreases the chance of inferring multiple paternity associated with relatively rare alleles.333
In this study we chose to maximize discrimination of multiple paternity at the expense of334
including higher levels of potential null alleles. Previous studies in this system using the335
same markers detected null alleles at a frequency of 1.8% (Tennessen and Zamudio, 2003).336
337
Computer Simulations338
We performed an initial set of simulations to test the effect of sampling regimes on339
detection of multiple paternity in this population of Ambystoma maculatum. Not340
surprisingly, BROOD simulations suggest that the number of sampled offspring necessary341
to detect multiple paternity increases as the number of contributing males increases (e.g.342
number of sampled offspring required increases from <10 larvae with one contributing343
father to 65 when 8 males contribute to the clutch). We also simulated various sampling344
conditions to evaluate our ability to detect various levels of multiple paternity in345
HAPLOTYPES (Fig. 2). Within clutch sampling has a greater effect on haplotype numbers346
than the size of the original clutch. Thus, small within clutch sampling limits the ability to347
discern the number of contributing males especially when three or more males sire348
offspring. Therefore the extent of multiple paternity could easily be underestimated when349
within clutch sample size is small. These results of subsampling simulations are in full350
concordance with those reported in the original description of the programs (refs).351
Given our BROOD results, clutches A, B, C, E, and F are sufficiently large for the352
detection of multiple paternity by at least two males, and clutches A, B, and C were large353
enough to detect multiple paternity by up to four or six males. Not surprisingly, these are354
the largest clutch sizes with 51, 32, and 46 genotyped offspring respectively. We chose355
Myers and Zamudio, Page 17
clutches A, B, and C for further analysis. Using the COUNTS_MED program we356
determined the number of unique haplotypes from unshared parents. Haplotype counts do357
not include loci with null alleles and therefore analysis was based on 7 loci for clutch A358
(excluding Ama 4-10, Ama A, and Ama 34), 6 loci for clutch B (excluding Ama 5-1, Ama359
A, Ama 12-7, and Ama 34), and 7 loci for clutch C (excluding Ama 61, Ama 4-10, and360
Ama 34). The number of unique haplotypes in clutches A, B, and C estimated by361
COUNTS were 44, 13, and 38, respectively. We used these numbers for simulations in the362
HAPLOTYPES program.363
First we performed simulations assuming no reproductive skew, by employing the364
programs as they were originally developed. We simulated three clutches (A, B and C) in365
HAPLOTYPES, each with the corresponding set of loci that were used in the COUNTS366
program. For clutch B, we used an approximation of 50 eggs with 20 sampled and367
estimated number of haplotypes and fathers for the simulated clutch. COUNTS determined368
that clutch B contained 13 unique haplotypes based exclusively on the genotypes of the369
offspring. Comparing this number to the simulated distribution of haplotypes, we can infer370
that this clutch was sired by a mean of 1.7 male parents (Table 2). However, one sire371
occurred twice as frequently as two sires in our simulations (Fig. 1); 95% confidence limits372
suggest that anywhere between one and four males may have contributed to this clutch.373
Thus, with only simulations, we cannot rule out single paternity or siring by two or more374
males. However, when simulation data are combined with the results from the Minimum375
Method, multiple paternity is confirmed and most likely consisted of two or three males376
(Table 2).377
Myers and Zamudio, Page 18
We simulated clutches A and C using a 100-egg clutch of which 50 were sampled.378
For clutch C, COUNTS determined 38 unique haplotypes. In the HAPLOTYPES379
simulation this corresponds to a mean of 4.1 males and the most common number of380
fathers for that haplotype count was 3 (Table 2; Fig. 1). Clutch A produced similar results.381
COUNTS determined 44 haplotypes corresponding to a mean of 5.9 sires and the most382
common number of males contributing to this clutch was 8 (Table 2; Fig. 1).383
In a second set of simulations we incorporated scenarios of reproductive skew to384
assess the impact of differential reproductive success on our estimates of multiple385
paternity. In BROOD simulations, differential male reproductive success within a clutch386
(increasing from no skew, to 2X, to 60%) dramatically increases N*, especially when a387
large number of males contribute to a clutch. For example, for a clutch with five388
contributing sires, the number of sampled larval genotypes required to detect multiple389
paternity are 40, 46, and 74 respectively for increasing skew scenarios. We also simulated390
the probability of detecting multiple paternity using a geometric distribution for391
contributions of multiple sires (Table 3) and in most instances our program was unable to392
compute subsampling because it required more offspring to find N* than were present in393
the clutch. Thus, multiple paternity could easily be undetected if reproductive skew is394
large.395
Increasing reproductive skew from an even distribution, to 2X, to the 60% scenario,396
generally increased the estimates of mean and most frequent number of contributing males397
(Table 2); however, incorporating reproductive skew in simulations also dramatically398
decreases the confidence in estimates of number of contributing males. Modeling399
reproductive skew as a geometric allocation of paternity in a female’s clutch results in a400
Myers and Zamudio, Page 19
less predictable pattern: the mean, and the most common number of simulated fathers401
increases (as in clutch B) or decreases (as in clutches A and C). Differences between the402
2x/60% results and the geometric distribution results are due in part to differences in their403
fundamental approach. The geometric distribution approach allows fathers to sire a certain404
proportion of available eggs in series (as would be expected in systems with serial mating)405
while the other methods account for a proportion of total offspring sired which may be406
more applicable in cases of sperm storage. Regardless of the model for reproductive skew,407
changing male representation within a clutch in HAPLOTYPES simulations produced a408
dramatic shift in the frequency curves resulting in a flatter or evenly parceled distribution409
curves for the estimated number of males contributing to a clutch (Fig 2).410
411
Discussion412
Multiple paternity413
Studies of a variety of taxa have demonstrated that multiple mating and multiple414
paternity can be important components of species-specific reproductive strategies415
(Birkhead 2000). In this study, we have confirmed that multiple paternity occurs416
commonly in natural spotted salamander breeding aggregations. Multiple paternity by 2-3417
sires was evident in more than 70% of egg masses using the conservative Minimum418
Method; therefore, it is probable that multiple paternity may be even more common. Our419
analyses also suggest that the number of sires contributing a clutch can be relatively high,420
with as many as eight males contributing to a clutch. These results have important421
implications for future studies of the costs and benefits of amphibian reproductive422
Myers and Zamudio, Page 20
strategies. Such a high prevalence of multiple paternity suggests a selective advantage for423
multiple mating and increased male representation in a female’s brood.424
The spotted salamander is a good study species to address questions about the425
benefits underlying the evolution and maintenance of this mating system. For females, the426
benefits of multiple mating come in both material and genetic forms. However, it is427
unlikely that material benefits are a factor influencing the evolution and maintenance of428
this mating system; unlike insects (Thornhill 1983), female salamanders do not receive429
nutrients from either the male’s spermatophores (Arnold 1976; Petranka 1998) or from430
nuptial gifts. In addition, females do not benefit from parental care of eggs or larvae431
because clutches are left unattended after oviposition. Lastly, material benefits do not432
include sperm quantity in this system. Studies of the reproductive biology of salamanders433
suggest that females are not sperm limited and could fertilize and deposit a complete clutch434
of eggs with a single or few spermatophores (Halliday & Verrell 1984; Sever et al. 1999).435
A direct assessment of all the possible genetic benefits for multiple mating is436
beyond the scope of this study. However, given our knowledge of the breeding behavior437
and natural history of this species, we can briefly evaluate several of the proposed438
hypotheses, including intrinsic male quality, offspring diversity, and the genetic439
compatibility hypotheses (Newcomer et al. 1999). The ‘intrinsic male quality’ hypothesis440
(Zeh & Zeh 2001) assumes that multiple mating enables sperm competition and cryptic441
female choice, thus increasing the probability that the highest quality sperm fertilizes eggs.442
This is consistent with ambystomatid breeding tactics where the female acquires many443
spermatophores thus providing an opportunity for sperm mixing and competition. The444
male reproductive skew we observe in our clutches lends support to this hypothesis; it is445
Myers and Zamudio, Page 21
possible the highest quality male fertilizes a majority of the clutch while secondary males,446
with lower quality or poorly competing sperm, father fewer offspring. However, this447
pattern of reproductive skew may simply be a result of male precedence rather than quality448
differences in sperm from different males (Jones et al. 2002). Offspring diversity and449
genetic compatibility may also play a role in this system. Genetic compatibility has450
previously been shown to be a significant factor influencing mating success in salamanders451
(Garner & Schmidt 2003). Variable environmental conditions leading to initial spring452
breeding can be highly variable even at the local level so a diverse clutch of offspring may453
be a way for females to hedge their bets (Thompson & Gates 1982, Skelly 2003).454
455
Detecting multiple paternity in natural populations456
In a highly fecund species with large numbers of offspring, it is rarely possible to457
genotype all of the offspring from a single reproductive bout (DeWoody et al. 2000).458
Consequently, computer simulations that take into account population allele frequencies459
and reproductive skew are critical to estimate the number of offspring needed to detect460
multiple paternity. In the spotted salamander, egg mortality can be high and this loss of461
larvae further compromises sample sizes. BROOD determined that in the cases of highest462
clutch mortality in our study, the remaining larvae, even if fully genotyped, were often463
insufficient in number to establish multiple paternity. Methods to reduce embryo mortality464
in the lab or extract DNA from early stage embryos would thus enhance our ability to465
characterize reproductive patterns in this species.466
Our results emphasize the importance of accounting for reproductive skew in467
estimates of multiple paternity when parental genotypes are unknown. For most systems,468
Myers and Zamudio, Page 22
we rarely have a priori knowledge of the degree of reproductive skew among males469
contributing to a clutch of eggs; in fact, this is often one of the parameters we wish to470
estimate in natural populations. Our simulations emphasize that increases in reproductive471
skew decrease our power to determine the number of males contributing to a clutch. Thus,472
one fruitful strategy might be to complement paternity analyses of wild-collected clutches473
with studies of clutches resulting from controlled matings where paternal contributions are474
known (Tennessen & Zamudio 2003). This approach allows us to limit the number of475
contributing males and more accurately estimate the differences in male reproductive476
success. Likewise, when the possible range of reproductive skew is known, as was the case477
in our study (Tennessen and Zamudio, 2003), it is still important to simulate various skew478
scenarios to infer a range of potential contributing males rather than rely on the simplistic479
assumptions of equal contribution by competing sires.480
481
Acknowledgments482
We thank A. M. Wieczorek and G. Clark for microsatellite development and483
optimization. We thank C. Chandler, J. Donolley, S. Hedtke, C. Jennings, J. Ortega, D.484
Reid, J. Tennessen, and C.Voyer for assistance in the field or lab; R. Bowden, L. Chan, K.485
Galbreath, H. Greene, D. Harvell, F. Janzen, J. Robertson, F. Rousset, M. Shulman, and486
three anonymous reviewers for constructive comments on the manuscript; and A. Fiumera,487
A. DeWoody, and Y. DeWoody for advice on computer programs. C. Brunner, the Cornell488
CubeSat team, and D. Adams allowed us to use their computers for simulations. This489
research was funded in part by a Morley Student Research Grant, an Einhorn Discovery490
Grant, and a grant from the Cornell Undergraduate Research Program (to EMM) and NSF491
Myers and Zamudio, Page 23
Grant DEB9907798 and funds from the College of Arts and Sciences, Cornell University492
(to KZ).493
Myers and Zamudio, Page 24
References494
Arnold SJ (1976) Sexual behavior, sexual interference, and sexual defense in the495
salamanders Ambystoma maculatum, Ambystoma tigrinum, and Plethodon jordani496
Zeitschrift für Tierpsychologie 42, 247-300.497
Birkhead T (2000) Promiscuity: an evolutionary history of sperm competition. Faber and498
Faber Ltd., London.499
Bishop SC (1941) The Salamanders of New York. The University of the State of New500
York, Albany.501
Denoel M, Poncin P, Ruwet JC (2001) Alternative mating tactics in the alpine newt502
Triturus alpestris alpestris Journal of Herpetology 35, 62-67.503
DeWoody JA, DeWoody YD, Fiumera AC, Avise JC (2000) On the number of504
reproductives contributing to a half-sib progeny array Genetical Research 75, 95-505
105.506
Dowling TE, Morit C, Palmer JD, and Rieseberg LH (1996) Nucleic acids III: Analysis of507
fragments and restriction sites. In: Molecular Systematics. (eds. Hillis DM, Moritz508
C, Mable BK), pp 249-320. Sinauer Associates, Inc, Sunderland, Massachusetts.509
Emlen ST, Oring L (1977) Ecology, sexual selection, and the evolution of mating systems510
Science 146, 215-223.511
Fitzsimmons NN (1998) Single paternity of clutches and sperm storage in the promiscuous512
green turtle Chelonia mydas Molecular Ecology 7, 575-584.513
Fiumera AC, DeWoody YD, DeWoody JA, Asmussen MA, Avise JC (2001) Accuracy and514
precision of methods to estimate the number of parents contributing to a half-sib515
progeny array Journal of Heredity 92, 120-126.516
Myers and Zamudio, Page 25
Flageole S, Leclair R Jr (1992) Étude démographique d’une population de salamanders517
(Ambystoma maculatum) à l’aide de la méthode squeletto-chronologique Canadian518
Journal of Zoology 70, 740-749.519
Garner TWJ, Schmidt BR (2003) Relatedness, body size and paternity in the alpine newt,520
Triturus alpestris Proceedings of the Royal Society of London, B 270,619-624.521
Haddad CFB, Pombal JP (1998) Redescription of Physalaemus spiniger (Anura:522
Leptodactylidae) and descriptions of two new reproductive modes Journal of523
Herpetology 32, 557-565.524
Halliday TR, Verrell PA (1984) Sperm competition in amphibians. In: Sperm Competition525
and the Evolution of Animal Mating Systems (ed. Smith RL), pp. 487-508.526
Academic Press, Inc, Orlando.527
Hillis DM (1977) Sex ratio, mortality rate, and breeding stimuli in a Maryland population528
of Ambystoma maculatum Bulletin of the Maryland Herpetological Society 13, 84-529
91.530
Husting EL (1965) Survival and breeding structure in a population of Ambystoma531
maculatum Copeia 1965, 352-362.532
Jones AG, Adams EM, Arnold SJ (2002) Topping off: A mechanism of first-male sperm533
precedence in a vertebrate Proceedings of the National Academy of Sciences USA534
99, 2078-2081.535
MacWhirter, RB (1991) Effects of reproduction on activity and foraging behavior of adult536
female Columbian ground squirrels Canadian Journal of Zoology 69, 2209-2216.537
Myers and Zamudio, Page 26
Madsen T, Olsson M, Wittzell H, et al. (2000) Population size and genetic diversity in sand538
lizards (Lacerta agilis) and adders (Vipera berus) Biological Conservation 94, 257-539
262.540
Metts B (2001) Ambystoma maculatum: Reproduction Herpetological Review 32, 98.541
Møller, AP (1998) Sperm competition and sexual selection. In: Sperm competition and542
sexual selection (eds. Birkhead TM, Møller AP), pp. 55-90. Academic Press,543
London.544
Newcomer SD, Zeh JA, Zeh DW (1999) Genetic benefits enhance the reproductive success545
of polyandrous females Proceedings of the National Academy of Sciences of the546
USA 96, 10236-10241.547
Nunney, L (1993) The influence of mating system and overlapping generations on548
effective population size Evolution 47, 1329-1341.549
Olsson M, Madsen T (2001) Promiscuity in sand lizards (Lacerta agilis) and adder snakes550
(Vipera berus): causes and consequences Journal of Heredity 92,190-197.551
Petranka JW (1998) Salamanders of the United States and Canada. Smithsonian552
Institution Press, Washington D.C.553
Roberts JD, Standish RJ, Byrne PG, and Doughty P (1999) Synchronous polyandry and554
multiple paternity in the frog Crinia georgina (Anura: Myobatrachidae) Animal555
Behaviour 57, 521-526.556
Roberts WE (1994) Explosive breeding aggregation and parachuting in a neotropical frog557
Agalychnis saltator (Hylidae) Journal of Herpetology 28, 193-199.558
Roff DA (2002) Life History Evolution. Sinauer Associates, Inc, Sunderland.559
Myers and Zamudio, Page 27
Rowe CL, Dunson WA (1995) Impacts of hydroperiods on growth and survival of larval560
amphibians in temporary ponds of central Pennsylvania Oecologia 102, 397-403.561
Sambrook J, Russell DW (2001) Molecular Cloning: a Laboratory Manual, 3rd edn. Cold562
Spring Harbor Laboratory Press, New York.563
Savage WK, Zamudio KR (2003) Ambystoma maculatum. In: Status and Conservation of564
US Amphibians Volume 2: Species Accounts (ed. Lannoo MJ). University of565
California Press, Berkeley (in press).566
Sever DM, Halliday TR, Waights V, Brown J, Davies HA, Moriarty EC (1999) Sperm567
storage in females of the smooth newt (Triturus v. vulgaris L.): I. ultrastructure of568
the spermathecae during the breeding season. Journal of Experimental Zoology569
283, 51-70.570
Skelly DK (2003) Microgeographic countergradient variation in the wood frog (Rana571
sylvatica). Evolution (in press).572
Stenhouse SL (1987) Egg mortality and recruitment of juveniles of Ambystoma maculatum573
and Ambystoma opacum in North Carolina Herpetologica 43, 496-501.574
Tennessen JA, Zamudio KR (2003) Early-male advantage, multiple paternity, and sperm575
storage in an amphibian aggregate breeder Molecular Ecology (in press).576
Thompson EL, Gates JE (1982). Breeding pool segregation by the mole salamanders,577
Ambystoma jeffersonianum and A. maculatum, in a region of sympatry Oikos 38,578
273-279.579
Thornhill R, Alcock J (1983) The evolution of insect mating systems. Harvard University580
Press, Cambridge.581
Myers and Zamudio, Page 28
Tregenza T, Wedell N (2002) Polyandrous females avoid costs of inbreeding Nature 415,582
71-73.583
Verrell PA, Pelton, J (1996) The sexual strategy of the central long-toed salamander,584
Ambystoma macrodactylum columbianum, in south-eastern Washington Journal of585
Zoology 240, 37-50.586
Verrell PA (1989) Sexual strategies of natural populations of newts and salamanders587
Herpetologica 45, 265-282.588
Wells KD (1977) The social behavior of anuran amphibians Animal Behaviour 25, 666-589
693.590
Whitford WG, Vinegar A (1966) Homing, survivorship, and overwintering of larvae in591
spotted salamanders, Ambystoma maculatum Copeia 1966, 515-519.592
Wieczorek AM, Zamudio KR, King TL, and Gajetvaj B (2002) Isolation of microsatellite593
loci in spotted salamanders (Ambystoma maculatum) Molecular Ecology Notes 2,594
313-315.595
Willmott HE, Foster, SA (1995) The effects of rival male interaction on courtship and596
parental care in the fourspine stickleback, Apeltes quadracus Behaviour 132, 997-597
1010.598
Zeh JA, Zeh DW (2001) Reproductive mode and the genetic benefits of polyandry Animal599
Behaviour 61, 1051-1063.600
Myers and Zamudio, Page 29
Author Information Box601
This research was conducted in partial fulfillment of Erin Myers’ undergraduate602
honors thesis. Research in the Zamudio lab focuses on population differentiation and603
mating system evolution in amphibians and reptiles. We are particularly interested in the604
determinants of reproductive success in aggregate breeding species.605
Myers and Zamudio, Page 30
Figure Legends606
Fig. 1. Results of simulations in the HAPLOTYPES program for clutches A, B, and607
C, assuming equal contribution of multiple males to eggs in a clutch (no reproductive608
skew). Histograms on the left reflect the number of times that 44, 13, and 38 haplotypes609
(the empirical number for each of those clutches estimated by COUNTS) were seen in610
simulations assuming no reproductive skew. Curves on the right represent the number of611
contributing fathers inferred given increasing numbers of observed haplotypes among612
offspring. White circles indicate the mean estimate of male parents, black circles indicate613
the number of male parents most commonly recovered in simulations, and grey circles are614
cases where the mean and most common number of sires are equal. Black triangles above615
each curve represent the number of haplotypes estimated for each clutch by COUNTS.616
Error bars represent the 95% confidence intervals about the mean (rounded to the nearest617
integer). Note that the maximum number of haplotypes possible in a given simulation is618
the same as the number of eggs sampled, therefore the simulation for clutch B includes a619
smaller range of possible haplotypes.620
Fig. 2: The effect of incorporating reproductive skew in HAPLOTYPES621
simulations on estimates of multiple paternity. For clutches A (top row), B (middle row),622
and C (bottom row) histograms reflect the number of times that 44, 13, and 38 haplotypes623
(the empirical number for each of those clutches estimated by COUNTS) were observed624
for varying numbers of sires in simulations including reproductive skew (2X, 60% and 4625
points along the geometric distribution, a= 0.125, 0.3, 0.6, and 0.9). Overall counts for the626
histogram will vary because of differences in the number of times that a particular627
haplotype number was seen in 1000 simulations.628
Myers and Zamudio, Page 31
Table 1 Summary of genotype results and allelic distributions for microsatellite loci in629
seven clutches of Ambystoma maculatum. For each clutch (A-G) we list number of630
genotyped larvae (N), the alleles at each locus (in base pairs) and their frequency (in631
parentheses). Colors/outlines represent patterns inferred from genotypes for that particular632
locus using the minimum method: non-informative (plain text), multiple paternity (white633
box), null allele (gray shading), multiple paternity with null allele (gray shaded box), and634
multiple maternity (light gray shading).635
Clutch N Ama Ama Ama Ama Ama Ama Ama Ama Ama Ama61 5-1 9-4 11-2B 4-10 A 3-3 2C2 12-7 34
A 51 229(30)
384 (2) 205 (20) 237(27)
237(17)
133(10)
169(26)
207 (5) 294(18)
100(30)
251(32)
386(57)
217 (8) 243 (3) 241(32)
157(19)
187(44)
213(33)
304(62)
108(25)
253(32)
388 (1) 221 (39) 245 (1) 267(40)
159 (1) 237 (7) 215 (5) 310 (8) 98 (8)
390(13)
223 (28) 247(63)
235 (3) 169(44)
193 (1) 221(49)
92 (2)
392 (9) 207 (1) 175 (3) 90 (2)396 (9) 193 (5) 120 (4)398 (9) 86 (2)
B 32 229(16)
386 (3) 221 (58) 237(28)
243(56)
133 (2) 169(29)
207 (8) 300(10)
90 (3)
388 (3) 243 (2) 167 (8) 187(27)
213(35)
304(40)
98 (2)
390(16)
245 (1) 169(59)
189 (3) 221(11)
310(10)
106 (19
392(25)
247(19)
237 (1) 332 (2) 108(12)
394 (3) 241 (4) 118 (1)398 (6) 247 (3) 120 (1)400 (2) 245 (2)
C 46 229(17)
386 (3) 205 (38) 237 (1) 229 (2) 157 (1) 169(69)
207(41)
294 (1) 86 (3)
251(25)
390 (2) 221 (26) 243(32)
235(23)
169(89)
187(19)
213(35)
300 (7) 100 (6)
253(50)
392 (2) 225 (18) 247(31)
241(41)
173 (1) 221 (1) 215(10)
304(62)
102(10)
396(16)
243 (5) 255 (1) 267(12)
175 (1) 241 (1) 221 (3) 310(19)
108 (6)
398(67)
257(22)
112(14)118(14)
Myers and Zamudio, Page 32
Clutch N Ama Ama Ama Ama Ama Ama Ama Ama Ama Ama61 5-1 9-4 11-2B 4-10 A 3-3 2C2 12-7 34
D 7 251 (4) 384 (3) 221 (4) 243 (1) 243 (6) 133 (4) 169 (4) 207 (7) 304(14)
102 (1)
253(10)
386 (3) 223 (8) 247 (9) 263 (2) 157 (2) 187 (1) 209 (2) 108 (5)
388 (1) 235 (2) 257 (4) 267 (6) 169 (8) 193 (4) 213 (3) 118 (8)390 (1) 241 (3) 221 (2)396 (2)398 (4)
E 14 229 (6) 386 (6) 191 (4) 243 (2) 237 (4) 157 (8) 169(13)
207 (9) 294 (1) 90 (10)
251 (5) 388 (1) 205 (10) 245 (2) 243 (9) 169(12)
187 (5) 211 (2) 300 (7) 100 (1)
253(11)
390 (3) 217 (5) 247 (9) 265 (1) 193 (5) 213(14)
304(14)
102 (1)
392 (2) 221 (6) 255 (8) 267 (2) 195 (1) 215 (2) 330 (2) 110 (1)394 (2) 223 (2) 257 (5) 221 (2) 237 (1) 221 (1) 118
(12)398 (8) 249 (2) 120 (1)400 (2)
F 18 229 (5) 384 (1) 191 (13) 241 (1) 237 (2) 157 (2) 169(12)
195 (2) 294 (4) 90 (4)
251 (3) 388 (2) 211 (2) 243(15)
243(14)
167 (4) 187 (2) 207 (4) 304(11)
98 (1)
253 (8) 390 (6) 217 (2) 247 (8) 245 (2) 169 (3) 191 (1) 213(15)
324 (2) 106 (1)
392 (1) 219 (6) 263 (2) 175 (1) 193 (4) 221 (3) 334 (2) 108(13)
394 (2) 221 (10) 241 (1) 223 (4) 118 (3)398 (3) 227 (2)400 (1)
G 7 251 (6) 384 (2) 191 (1) 247(10)
237 (6) 169 (4) 169 (5) 207 (4) 300 (1) 90 (5)
253 (8) 386 (4) 205 (4) 243 (3) 243 (6) 167 (6) 187 (5) 213 (3) 304 (8) 98 (4)390 (6) 221 (7) 257 (1) 215 (2) 310 (3) 118 (5)392 (2) 223 (2) 221 (5)
Legend: Non-informative locus(single or multiplepaternity possible)
Null allele detected Multiple paternity Multiple paternitywith null allele
Multiple maternity
636
Tabl
e 2.
Esti
mat
es o
f mal
e co
ntrib
utio
n to
7 fi
eld-
colle
cted
clu
tche
s of A
mby
stom
a m
acul
atum
usin
g th
e M
inim
um M
etho
d an
d
HA
PLO
TYPE
S sim
ulat
ions
. Clu
tche
s A, B
, and
C m
et m
inim
um sa
mpl
e siz
es fo
r det
ectio
n of
mul
tiple
pat
erna
l alle
les;
we
repo
rt th
e
mea
n an
d m
ost f
requ
ent (
in p
aren
thes
es) n
umbe
rs o
f mal
es c
ontri
butin
g to
eac
h cl
utch
bas
ed o
n 10
00 si
mul
atio
ns u
nder
four
scen
ario
s
of m
ale
repr
oduc
tive
succ
ess:
no sk
ew ,
one
dom
inan
t mal
e w
ith tw
ice
the
num
ber o
f offs
prin
g as
oth
er m
ales
(2X
), on
e do
min
ant
mal
e w
ith 6
0% o
f offs
prin
g (6
0%) a
nd se
ven
leve
ls of
repr
oduc
tive
skew
sam
pled
from
a g
eom
etric
dist
ribut
ion
of p
ater
nity
with
in a
fem
ale’
s clu
tch.
Clu
tche
s D, E
, F, a
nd G
wer
e no
t inc
lude
d in
sim
ulat
ion
anal
yses
bec
ause
they
did
not
mee
t min
imum
sam
ple
size
requ
irem
ents
for d
etec
tion
of a
ll pa
tern
al a
llele
s or b
ecau
se o
f evi
denc
e of
mul
tiple
mat
erni
ty (M
M).
geom
etri
c dist
ribu
tion
Clu
tch
Min
imum
Met
hod
No
Skew
2X60
%a
=0.1
25
a=0
.2a
=0.3
a=0
.4a
=0.5
a=0
.6a
=0.7
a=0
.8a
=0.9
A3
5.9
(8)
6.3
(8)
6.1
(8)
6.8
(6)
6.5
(7)
6 (8
)5.
8 (5
)5.
5 (8
)5.
5 (7
)5.
3 (7
)5
(8)
4.2(
6)B
31.
7 (1
)2
(1)
3.9
(1)
3 (3
)2.
3 (2
)1.
9 (1
)2.
7 (1
)3.
5 (1
)4
(1)
4.2
(1)
4.4
(1)
4.6
(7)
C2
4.1
(3)
4.8
(4)
5 (8
)5.
4 (5
)5.
1 (4
)5.
1 (6
)5
(3)
5 (7
)4.
9(7)
4.8
(3)
4.5
(4)
4.7
(3)
D1
——
——
——
——
——
——
EM
M—
——
——
——
——
——
—F
3—
——
——
——
——
——
—G
2—
——
——
——
——
——
Myers and Zamudio, Page 33
Tab
le 3
. T
he e
ffec
t of
incl
udin
g re
prod
uctiv
e sk
ew in
BR
OO
D s
imul
atio
ns.
We
estim
ated
the
mea
n nu
mbe
r of
gen
otyp
ed o
ffsp
ring
requ
ired
to d
etec
t alle
lic f
requ
enci
es o
f m
ultip
le p
aren
ts f
or a
clu
tch
of 1
00 e
ggs
unde
r fo
ur s
cena
rios
of
repr
oduc
tive
skew
: i)
no
skew
, ii)
a d
omin
ant m
ale
with
2X
mor
e of
fspr
ing
then
oth
er m
ales
con
trib
utin
g to
the
clut
ch, i
ii) a
dom
inan
t mal
e w
ith 6
0% o
f th
e
offs
prin
g in
a c
lutc
h, a
nd iv
) sk
ew m
odel
ed a
s a
geom
etri
c di
stri
butio
n w
ith v
aryi
ng d
egre
es o
f a (
prop
ortio
n of
rem
aini
ng e
ggs
sire
d
by e
ach
succ
essi
ve m
ale)
. In
som
e si
mul
atio
ns, t
he n
umbe
r of
gen
otyp
es r
equi
red
exce
eded
the
num
ber
of e
ggs
in o
ur s
imul
ated
clut
ch (
case
s m
arke
d w
ith a
n as
teri
sk).
geom
etri
c di
stri
butio
n
# m
ales
no s
kew
2X60
%a
=0.1
25a
=0.2
a=0
.3a
=0.4
a=0
.5a
=0.6
a=0
.7a
=0.8
a=0
.9
14
--
--
--
--
--
-
213
1614
1313
1415
1619
2329
51
322
2632
2223
2728
3544
65*
*
431
3651
3034
4149
6477
**
*
540
4674
3950
5778
**
**
*
650
5589
4760
73*
**
**
*
757
6390
5670
82*
**
**
*
866
7210
064
78*
**
**
**
Myers and Zamudio, Page 34
number of contributing males
0 20 40 60
2
4
6
8
01 2 3 4 5 6 7 8
200
160
120
80
40
0
clutch A
number of haplotypes
1 2 3 4 5 6 7 8
100
80
60
40
20
0
clutch B
1 2 3 4 5 6 7 8
100
80
60
40
20
0
clutch C
nu
mb
er o
f co
ntr
ibu
tin
g m
ales
cou
nt
Myers and Zamudio, Page 35
0 20 40 60
2
4
6
8
0
0 20 40 60
2
4
6
8
0
Figure 1
Figure 2
Myers and Zamudio, Page 36
���
������������ �������
�����
��
��
�
��
�
��
�
��
��
��
��
��
��
��
��
��
��
�
��
�
��
� �
��
�
��
�
����
�����
��
��
�α
= 0
.125
α =
0.3
α =
0.6
α =
0.9
Myers and Zamudio, Page 37
Appendix. Allele frequencies for 10 microsatellite loci in a random sample of adult637Ambystoma maculatum from Ringwood Pond, Tompkins Co., New York. These638frequencies were used in the HAPLOTYPE simulations to determine the number of639contributing sires to field-collected clutches.640
641Locus Allele Frequency Locus Allele FrequencyAma61 229 0.085 Ama3-3 169 0.031
251 0.272 173 0.008253 0.638 179 0.004255 0.004 180 0.004
183 0.008Ama5-1 384 0.003 184 0.004
386 0.234 186 0.004388 0.011 187 0.343390 0.136 189 0.038392 0.250 193 0.008394 0.008 213 0.008396 0.043 217 0.004398 0.313 219 0.013400 0.003 221 0.021
229 0.008Ama9-4 191 0.002 233 0.025
205 0.167 237 0.025207 0.070 241 0.017213 0.004 245 0.004215 0.004 247 0.004217 0.044 249 0.017219 0.004 251 0.008221 0.454 253 0.017223 0.118 255 0.013225 0.039 257 0.013227 0.009 259 0.004229 0.013 261 0.004235 0.022 263 0.025237 0.022 351 0.013241 0.004243 0.009 Ama34 86 0.002245 0.013 90 0.265
92 0.005Ama11-2B 237 0.155 94 0.005
243 0.077 98 0.010245 0.027 100 0.005247 0.591 106 0.010249 0.023 108 0.158251 0.005 110 0.040255 0.009 112 0.015257 0.091 116 0.228269 0.005 118 0.252271 0.005 120 0.005273 0.009275 0.005
Myers and Zamudio, Page 38
642Locus Allele Frequency Locus Allele Frequency
Ama4-10 235 0.052 Ama2C2 203 0.018237 0.157 207 0.215241 0.052 209 0.039243 0.313 211 0.031263 0.004 213 0.382265 0.009 215 0.035267 0.348 217 0.018269 0.065 221 0.224
223 0.031AmaA 133 0.002 229 0.009
151 0.025157 0.112 Ama12-7 236 0.004159 0.017 242 0.004163 0.013 290 0.004165 0.008 294 0.091167 0.033 300 0.161169 0.548 302 0.022171 0.200 304 0.624173 0.021 308 0.013175 0.004 310 0.074193 0.004 330 0.002197 0.013
643
Top Related